Article title: Negation in Chinese: A corpus-based study

Authors: Richard Xiao & Tony McEnery

Affiliation: Lancaster University

Address: Lingusitics Department

Lancaster University

Bailrigg, Lancaster

Lancashire

LA1 4YT

United Kingdom

Telephone: +44 1524 593025

Fax: +44 1524 843085

Email: z.xiao@lancaster.ac.uk

Negation in Chinese: A corpus-based study

This article explores negation in Chinese on the basis of written and

spoken corpora of Mandarin Chinese. The use of corpus data not only

reveals central tendencies in language based on quantitative data, it

also provides typical examples attested in authentic contexts. In this

study we will first discuss negation variants in general terms, on the

basis of which the focus will be shifted to the semantic and genre

distinctions between two major negators, bu and mei (including

meiyou). Following this is an exploration of the interaction between

negation and aspect marking. We will then move on to discuss scope

and focus of negation, transferred negation, and finally double

negation and redundant negation.

1. INTRODUCTION

While negation in Chinese has been subject to intensive study for decades,

many issues remain unresolved. For example, many competing and even

conflicting characterisations have been proposed of the distintions between bu

and mei/meiyou (hereafter referred to as mei unless otherwise stated, notably

in section 2); there is also little concensus over such issues as whether the

progressive can be negated by bu or mei; many generalisations about negation

found in the liteterature cannot account for attested language data; the more

fine-grained distinctions in the distribution of negators across genres are

largely unexplored. A common failure of those studies is that they have not

taken account of attested data, being based instead on a handful of examples

which are invented for linguistic analysis on the basis of the intuitions of

individual authors. Intuitions, however, can be biased and vary from person to

person (cf. cf. Seuren (1998, 260-262)). They are not always reliable, thus

accounting for the many conflicting characterisations common in the literature

(see section 3).

In this study we take a corpus-based approach to negation in Chinese. The

use of corpus data can not only reveal central tendencies in language, it can

also provide typical examples attested in authentic contexts. This is because a

corpus pulls together the intuitions of a large number of language users and

the corpus-based approach focuses on what is quantitatively central and

typical in attested language data. With that said, we do not claim that the

intuitions of individual researchers should be disregarded in linguistic analysis.

On the contrary, we seek to achieve a marriage between the corpus-based and

intuition-based approaches, as have been advocated in Xiao and McEnery

(2004) and McEnery, Xiao and Tono (2005).

The major corpus resources used in this study include the Lancaster

Corpus of Mandarin Chinese (LCMC) and the Callhome Mandarin Chinese

Transcript. LCMC is a balanced corpus which was designed as a Chinese

match for FLOB (see Hundt, Sand and Siemund (1998)), representing written

Chinese published in China in the early 1990s (see McEnery, Xiao and Mo

(2003)). It consists of five hundred 2,000-word samples taken proportionally

from 15 genres raning from news reportage to academic writing, totalling one

million words (see Table 1). The Callhome corpus, which was released by the

Linguistic Data Consortium (LDC) in 1996, comprises a contiguous 5 or 10

minute segment taken from 120 unscripted telephone conversations between

native speakers of Mandarin Chinese, totalling approximately 300,000 words.

In addition to these two corpora, we will occasionally consult two much larger

corpora, the PDC2000 and Mandarin News Text corpora, for very infrequent

features or to confirm what is observed in the LCMC and Callhome corpora.

The PDC2000 corpus covers one year’s newspaper texts published by the

People’s Daily in 2000, totalling approximately fifteen million words.1 The

Mandarin News Text Corpus, which was also released by the LDC, contains

over 200 million words of newspaper texts from the People’s Daily (1991-

1996), newswire texts from the Xinhua News Agency (1994-1996) and

transcripts from China Radio International (1995-1996). While these are

specialised corpora, they can nevertheless provide some supporting evidence

for our observations based on LCMC and Callhome.

[TABLE 1 NEAR HERE]

In the remainder of this article, we will first discuss negation variants in

general terms (section 2), on the basis of which the focus will be shifted to the

semantic and genre distinctions between two major negators, bu and mei

(section 3). Following this is an exploration of the interaction between

negation and aspect marking (section 4). We will then move on to discuss the

scope and focus of negation (section 5), transferred negation (section 6), and

finally double negation and redundant negation (section 7). Section 8

concludes the article by summarising the findings of this study.

2. NEGATION VARIANTS IN CHINESE

There are about a dozen negative adverbs in Chinese including, for example,

bie ‘don’t’, beng ‘don’t; needn’t’, bu ‘not’, fei ‘non-; not’, mei ‘no, not’, mo

‘no, not’, wei ‘not yet’, wu1 ‘no, not’, wu2 ‘do not’, wu3 ‘do not’, xiu ‘do not’

as well as some compound words consisting of them: bubi ‘need not’, buceng

‘not ever’, buyao ‘do not’, buyong ‘need not’, meiyou ‘no, not’, shangwei ‘not

yet’, weibi ‘not necessarily’, weiceng ‘not ever’, weichang ‘not ever; not

necessarily’, wuxu ‘need not’, and wuyong ‘need not’. These negative forms

have different functions. For example, beng, bie, buyao, mo, wu2, wu3 and xiu

are typically used in imperatives; fei occurs most frequently as a prefix or in

double negation structures (see section 7); bubi, buyong, wuxu, and wuyong

negate the necessity of doing something while buceng, shangwei, wei, weiceng,

and weichang (its first meaning above) are aspectually related, negating the

realisation of a situation. The most important negators in Chinese are bu and

mei, which will be discussed in detail in the following sections.

Of the six negative adverbs used in imperatives, buyao is most frequent

(21 and 92 per 100,000 words in LCMC and CallHome), followed by bie (16

and 51 instances per 100,000 words in the two corpora). The other adverbs are

very infrequent in both corpora. Beng was only found in the spoken corpus

(five instances per 100,000 words) whereas wu3 (one instance per 100,000

words) was only found in the written corpus; mo, wu2 and xiu are extremely

rare in both written and spoken Chinese. It is clear that negative adverbs are

significantly more common in spoken data, which is hardly surprising given

that imperatives typically occur in spoken discourse and dialogues in literary

works.

There are 409 instances of fei in LCMC, of which 321 instances (over

three quarters) appear as a prefix in technical terms (e.g. fei-lixing ‘non-

rational’, fei-zhengfu ‘non-government’). The remaining 88 instances typically

occur in the double negation structure fei…(bu/mo) ‘must, have to’ (see

section 7) or in formulaic expressions such as fei…fei ‘neither…nor’ and

fei…ji ‘either…or’. In contrast, while fei is infrequent in the CallHome spoken

corpus, the examples present differ from the LCMC findings. Only three out

of the 13 instances of fei in Callhome occurred as a prefix, with all others

found in double negation structures. Given that speech is relatively less

technical than writing (cf. Biber (1988)), it is quite unsurprising that the prefix

use of fei is significantly more frequent in written Chinese.

Of the adverbs negating necessity, buyong is most frequent (five and 51

instances per 100,000 words in LCMC and CallHome respectively). This also

counts as the only form used frequently in spoken Chinese if we disregard the

two occurrences of bubi. In contrast, more forms are used in written Chinese.

In addition to buyong and bubi (seven instances per 100,000 words), wuxu and

wuyong are found in written genres, though they are as infrequent, occurring

with a frequency of less than one instance per 100,000 words. The relative

frequencies across genres show that while all of these adverbs can negate

necessity, buyong is more colloquial, and bubi is more formal while both wuxu

and wuyong occur only in formal written genres.

Buceng, shangwei, wei, weiceng, and weichang negate the realisation of a

situation. They typically occur in written Chinese. The most common of these

is wei (28 instances per 100,000 words in LCMC), followed by shangwei

(seven instances per 100,000 words) while the other three are rare even in

written genres. The adverb that is used much more frequently to negate

realisation in both written and spoken Chinese is mei, which occurs nearly 200

times in 100,000 words, excluding its verb usage.

Mei and meiyou can be used either as adverbs or as verbs. When used as

verbs, they are the negative forms of you ‘have; there be’, which is a special

verb in Chinese that can only be negated by mei. As mei and meiyou are

interchangeable in most contexts in modern Chinese, the more subtle

distinctions between the two forms have rarely been noted in previous studies.

Historically mei appeared earlier than meiyou, with the adverbial use of

meiyou derived from its verb usage, i.e. the negative form of you ‘have; there

be’ (cf. Qu (2002, 2)). As such, mei tends to appear as an adverb (66% of the

time) while meiyou is more likely to used as a verb (57%) in written Chinese.

Interestingly, an opposite trend is observed in spoken Chinese, where mei is

slightly more likely to be used as a verb (53%) than as an adverb (47%) while

meiyou tends to appear as an adverb (62%) rather than as a verb (38%). The

contrast between written and spoken Chinese can be accounted for by two

important facts. On the one hand, the verb usage of mei is much more frequent

than that of meiyou in ‘economical’ expressions such as mei banfa ‘can do

nothing about it; can’t help it’ (with a mei/meiyou ratio of 3.7), mei guanxi ‘it

doesn’t matter, never mind’ (a ratio of 2.1), mei shi ‘it doesn’t matter; that’s

all right’ (a ratio of 4.1), and mei shi’r ‘it doesn’t matter; that’s all right’ (a

ratio of 12.5) while these expressions are much more common in speech than

writing (with a speech/writing ratio of 3.3). On the other hand, the adverbial

usage of meiyou is significantly more frequent than that of mei as a question

tag or an independent reply (over nine times as frequent in our data).2 These

two uses of mei and meiyou commonly occur in dialogues and are typical of

spoken Chinese. In fact, our data suggests that they are over eight times as

frequent in speech as in writing. In the written genres covered in LCMC, these

two uses were only found in five types of fiction type plus humour, which

resemble speech in many respects. A further difference between mei and

meiyou as an adverb is that mei is much more frequent (over ten times more

frequent in our data) in the V-mei/meiyou-V structure, which typically occurs

in a question or as the object clause of a verb phrase equivalent to ‘do not

know’, ‘be not sure’ and ‘find out’. While we acknowledge the differences

between mei and meiyou, we will use mei also to refer to meiyou for the sake

of conciseness.

Mei and bu are the most frequent and most important negators because

they generally reflect the features of the grammatical category of negation in

Chinese.3 Most negative adverbs discussed earlier can be replaced by bu and

mei (or their compounds), though the replacement may result in a stylistic

change. As bu is an adverb, we will only consider the adverbial usage of mei

in the remainder of this article so as to make the analysis more comparable.

3. DISTINCTIONS BETWEEN BU AND MEI

Both bu and mei typically negate verbs and adjectives.4 While the two adverbs

have been studied for decades, little has been agreed upon except that shi ‘be’

can only be negated by bu while you ‘have; there be’ can only be negated by

mei. In fact, many competing and even conflicting generalisations have been

proposed regarding the distinctions between bu and mei. For example,

according to Lü (1982), mei is used in objective narration with a past or

present reference time whereas bu negates subjective desire; Bai (2000) argues

that the major distinction does not lie in temporal references but between

objective narration and subjective evaluation instead; Li and Thompson (1981)

claim that bu is a neutral negative marker that denies the existence of a state

whereas mei denies the completion of an event/action; Hsieh (2001) and Lin

(2003) also argue that bu selects stative situations whereas mei selects

dynamic situations; according to Li (1999), bu is marked with [-telic] while

mei is marked with [+telic]; Shen (1995) views the distinction between bu and

mei as parallel to that between bounded and unbounded situations; and finally,

Liu (1997) argues that the two negative markers differ in mood, i.e. mei is

realis while bu is irrealis. Table 2 summarises the major distinctions which

have been proposed for bu and mei in the literature.

[TABLE 2 NEAR HERE]

While it might be possible to say that mei negates a situation objectively, it

is hard to see why the negation marked by bu should be viewed as subjective.

It is true that when bu modifies volitional verbs such as xiang ‘want to’ and

xiwang ‘hope’, what is negated is the subjective desire of the speaker; but such

a reading is impossible with non-volitional verbs such as cunzai ‘exist’ as

shown in (1a).5 Neither is it appropriate to say that mei cannot negate a future

situation. As can be seen from (1b), in relation to danxin ‘worry’, it is surely a

future situation to be able to reach a disarmament agreement, yet it is negated

by mei. Neng ‘can’ in this example is a modal verb, which is stative and has

nothing to do with aspectual features such as completion, telicity or

boundedness. Yet neng can be negated by mei. It might be argued that neng

‘can’ is a special verb, but is not uncommon to find counter examples for the

proposed distinctions in attested data. For example, duoshan ‘dodge’ is a

dynamic verb negated by bu in (1c); zhao lüguan ‘look for a hotel’ is an atelic

and unbounded situation which is irrelevant to completion. However, it is

equally plausible to say wo meiyou zhaodao lüguan ‘I did not find a hotel’. In

this modified example, zhaodao ‘find’ denotes a telic situation that can be

completed. It is clear that both zhao ‘look for’ and zhaodao ‘find’ can be

modified by mei. With an atelic situation, mei negates its realisation; with a

telic situation, mei negates the attainment of its final endpoint instead of its

realisation.

(1) a. shijie shang genben bu cunzai guiguai (LCMC: E)

world on at:all not exist ghost

‘No ghost exists in the world at all.’

b. renmen danxin, ruguo dao 5-yue 22-ri zhiqian hai mei neng

people worry, if by May 22 before still not can

jiu jiechu wuzhuang dacheng yizhi, Bei’ai heping

on remove arms reach agreement N.:Ireland peace

jincheng jiang zaodao zhiming de daji (PD 25/05/2000)

progress will suffer fatal DE blow

‘There are worries that if no agreement can be reached on

disarmment by May 22nd, the peace progress in Northern

Ireland will suffer from a fatal blow.’

c. ta benneng de ba yanjing yi bi, que ying-zhe

she instinct DE BA eye one close but toughen-Asp

toupi bu duoshan (LCMC: P)

scalp not dodge

‘She closed her eyes by instinct, but forced herself not to

dodge.’

d. wo meiyou zhao lüguan (LCMC: K)

I not look:for hotel

‘I did not look for a hotel.’

Nor is the difference between bu and mei parallel to the realis vs. irrealis

distinction. Liu (1997) argues that in the realis mood the declarative form

takes the verb-final -le or sentence-final le and the negative form is marked by

mei, while in the irrealis mood the declarative form takes hui ‘will’ and the

negative form is marked by bu(hui) ‘(will) not’, assuming that hui ‘will’ can

always be inserted after bu (as in buhui ‘will not’). This argument, nonetheless,

is untenable, because on the one hand, hui ‘will’ cannot be inserted, if at all,

after bu without a change in meaning, while on the other hand, the relationship

between -le and mei is asymmetrical. Indeed, all situations that can take -le

can be negated by mei, but the reverse is not true. A further weakness of this

theory is that it cannot account for the felicitous co-occurrence of bu and the

sentence-final le as in wo bu zou le ‘I will not go’ (LCMC: K), because a

situation cannot be both realis and irrealis simultaneously.

It can be seen even from this brief review that negation in Chinese is

closely associated with tense and aspect. Some of the explanations forwarded

for the phenomena observed so far can indeed account for some phenomena

observed in attested data, e.g. most situations negated by mei are dynamic and

have a past or present reference time. Nevertheless, the purely intuition-based

approach taken in these studies (e.g. Lü (1982); Li (1999); Hsieh (2001)) also

defines their limitations. None of the claims we have reviewed so far is

adequate in the light of our corpus evidence. A common failure of non-corpus-

based studies is that they tend to overlook what is obvious and typical in

attested language data selecting instead marginal examples, with such

examples being common in intuition based studies “because each of us has

only a partial knowledge of the language, we have prejudices and preferences,

our memory is weak, our imagination is powerful (so we can conceive of

possible contexts for the most implausible utterances), and we tend to notice

unusual words or structures but often overlook ordinary ones” (Krishnamurthy

(2000, 32-33)).

While previous studies have focused on aspectual features such as

dynamicity, telicity and boundedness which are associated with situation

aspect (see Xiao and McEnery (2004) for a discussion), a close inspection of

the concordances from the LCMC (6,733 instances of bu and 1,061 instances

of mei as an adverb) and Callhome (4,805 instances of bu and 2,180 instances

of mei as an adverb) corpora shows that mei negates the realisation of a

situation whereas bu is used in a much wider range of contexts to negate, for

example, possibility, necessity, property, status, volition, or simply the

truthfulness of judgment (i.e. the relationship between the subject and the

predicate), thus explaining why bu is much more frequent than mei.

The contrast between mei and bu in this respect is particularly marked in

cases where both negative adverbs are possible. As many authors have

observed, stative verbs like zhidao ‘know’, baohan ‘include’ and mingbai

‘understand’ are typically negated by bu. Indeed, as shown in example (2),

speakers may make a self correction from mei to bu in conversations.

Nevertheless, in some contexts these verbs can only be negated properly by

mei, contrary to the claim often found in the literature that they can never be

modified by mei. Consider, for example, the contrast between the (a) and (b)

sentences in (3-5). In (3a) and (4a) bu cannot be replaced with mei whereas in

(3b) and (4b) meiyou cannot be replaced with bu, even though the same verb is

used in the same pair of sentences. This is because the situations in (3a) and

(4a) have nothing to do with realisation whereas in (3b) and (4b) there is a

contrast between realisation (zhidao-le and baohan-le) and non-realisation

(meiyou zhidao and meiyou baohan). However, this contrast is not always

obvious. In (5a), for example, bu can possibly be replaced by mei, but the

replacement will result in a change in meaning: wo bu mingbai means that the

statement that I understood was not true, while wo mei mingbai means that the

situation wo mingbai did not occur. For this reason, meiyou is more

appropriate than bu in (5b). As (5b) expresses a series of events (mingbai

‘understand’, gen ‘follow’ and zou ‘walk’), a simple negation of prediction in

the first clause would be out of place. This contrast is also apparent in

dynamic verbs and adjectives. For example, in wo ji bu hejiu, ye bu chouyan ‘I

neither drink nor smoke’ (LCMC: R), what is negated is the speaker’s habit.

But if bu is replaced with mei, as in wo ji mei hejiu, ye mei chouyan ‘I neither

drank nor smoked’, then what is negated is the realisation of the events of

drinking and smoking on a particular occasion. Similarly, to answer the

question ni shi shou le haishi pang le? ‘Have you got fatter or slimmer?’

(Callhome), only mei is appropriate if a negative answer is given (compare

mei pang mei shou and bu pang bu shou).6 It is clear that in cases where both

bu and mei are possible, the choice of either generally entails a difference in

meaning (cf. also Nie (2001)).

(2) e, ta hai you ge jiali dianhua, wo haoxiang… jiali

uh, he also have Cl home telephone I seem home

haoxiang mei bu zhidao (Callhome)

seem not not know

‘Uh, he also has a home phone number, I seem… don’t seem to

know his home number.’

(3) a. shui ye bu zhidao ta de mingzi (LCMC: G)

who also not know she DE name

‘No one knows her name.’

b. ni zhidao-le wo de mingzi, wo ke hai meiyou zhidao

you know-Asp I DE name I but yet not know

ni de mingzi ne (LCMC: N)

you DE name Prt

‘You have known my name, but I haven’t known yours.’

(4) a. shishishang, guowai chengshu de, guifanhua de

in:fact abroad mature DE standardise DE

zhengquan shichang juedui bu baohan you ruci da

stock market definitely not include have so large

bizhong de guoyou gufen (LCMC: F)

proportion DE state-owned stock

‘In fact, the mature and standardised stock markets in foreign

countries definitely do not include such large proportions of

state-owned stocks.’

b. Beiyue de jianyi baohan-le Huayue de suoyou

NATO DE proposal include-Asp WTO DE all

junshi jidi, dan meiyou baohan Beiyue tebie shi Meiguo

military base but not include NATO especially be US

zai haiwai <…> de junshi jidi (LCMC: B)

in overseas DE military base

‘NATO’s proposal included all military bases of WTO (the

Warsaw Treaty Organization), but not those of NATO,

especially the overseas military bases of the United States.’

(5) a. dangshi wo bu mingbai ta weishenme zheyang jijiao,

at:that:time I not understand she why so haggle

jiu yansu de shuo <…> (LCMC: P)

then seriously DE say

‘At that time, I did not understand why she was so calculating,

so I said seriously…’

b. wo mei mingbai Hou de yishi, gen zai ta

I not understand Hou DE meaning follow in she

shenhou, xiang youlang jintou de yi shan men zou

body:behind toward veranda end DE one Cl door walk

qu (LCMC: P)

toward

‘I did not understand what Hou meant, so I followed her and

walked toward a door at the end of the veranda.’

Hsieh (2001, 26-27) produces two pieces of evidence to argue that the

difference between bu and mei cannot be the realised vs. unrealised distinction.

First, a clause negated by bu, just like one that takes mei, “can describe a

realised situation” because both can function as the complement of perception

verbs such as kanjian ‘see’, as shown in her examples wo kanjian ta yi dong ye

bu dong ‘I saw him not budge’ and wo kanjian ta mei(you) dong ‘I saw him

not move’. Hsieh assumes that the situations expressed by the complement

clauses of perception verbs must be realised since these verbs “denote direct

perception”; second, clauses negated by bu can function as the complement of

factive verbs such as yayi ‘surprised’ and yihan ‘regret’ as in her example wo

hen yayi ta bu hui youyong ‘I was surprised that he could not swim’. Hsieh

assumes that “[s]ince complements of factive verbs denote ‘realised’ situations,

the situation denoted by the bu complement should be considered as

‘realised’.”

This argument, nevertheless, is flawed in a number of ways. First, as noted

earlier, while mei negates the realisation of a situation, bu does not appear to

be related to realisation in any way; it does not negate an “unrealised”

situation as suggested. Second, neither bu nor mei can mark the realisation of a

situation. Perfective aspects are typically used to present a situation as it has

been realised or experienced, though covert aspect marking is also an

important strategy in Chinese discourse (cf. McEnery and Xiao (2002)). Mei,

in contrast, is used to negate realisation and experientiality (see section 4).

Finally, the two assumptions underlying this argument, i.e. the situations

expressed by the complement clauses of perception verbs and those denoted

by factive verbs are always realised, are totally ungrounded. Let us first

consider the following examples.

(6) a. zai lou wai de renmen kanjian yi chang

in building outside DE people see one Cl

sirenxinfei de beijü jiuyao fasheng le (LCMC: A)

piercing DE tragedy be:about:to happen COS

‘People outside the building saw that a piercing tragedy was

impending.’

b. di’er tian, qizi you jian zhangfu zai jiyan (LCMC: R)

second day wife again see husband Asp smoke

‘The next day, the wife saw her husband smoking again.’

c. wo hen qiyi shenme ren zai jiao wo ne (LCMC: G)

I very surprised what person Asp call I Prt

‘I was very surprised that someone should be calling me.’

Sentences like these are very common in Chinese. They contain either a

perception verb (e.g. kanjian ‘see’, jian ‘see’) or a factive verb (e.g. qiyi ‘be

surprised’). In (6a) the tragedy had not happened, but there were signs that it

was about to happen. That was what those people saw. In (6b) and (6c) the

progressive aspect marker zai was used. As the progressive is an imperfective

viewpoint, the situations were presented as ongoing, or in progress, which

means that these situations were also not realised. Given that complement

clauses denoting unrealised situations go perfectly well with perception verbs,

it is hard to see why the complement clauses taking bu or mei should be

considered as realised as Hsieh suggests. With a factive verb such as know,

believe, be sorry or be surprised, the speaker presupposes the truth of the

proposition expressed in the complement clause. But truth must not be

conflated with realisation. In (6c), for example, the proposition shenme ren zai

jiao wo ‘someone is calling me’ was true, but this does not entail that the

situation was realised, as indicated by the progressive aspect marker zai.

The different targets of negation for bu and mei account for why bu

modifies adjectives, stative verbs and modal verbs more frequently whereas

mei modifies dynamic verbs more frequently, as illustrated in Figures 1-2. The

figures show the proportions of bu and mei used to modify adjectives,

adverbials, dynamic verbs, stative verbs, and modal verbs, and of those

occurring in imperatives, as a negative potential infix (as in zhao-bu-dao

‘cannot find’, as opposed to the positive potential infix de as in zhao-de-dao

‘can be found’), and as an independent reply or a question tag. As can be seen,

in both LCMC and Callhome, bu modifies adjectives, stative verbs, and modal

verbs much more frequently whereas mei modifies dynamic verbs more

frequently. This is because situations that can be realised are mostly dynamic

while property, volition, relationship, possibility and necessity etc are all

stative. The distribution patterns of bu and mei are confirmed by their top ten

collocations, defined in this article as words immediately following bu and mei

which show a z-score greater than 3.0 and a minimum frequency of 3.

[FIGURE 1 NEAR HERE]

[FIGURE 2 NEAR HERE]

• Bu: gan ‘dare’, yuan ‘willing’, shi ‘be’, zhidao ‘know’, pingheng

‘balanced’, keneng ‘possible’, dong ‘understand’, qing ‘clear’, dong

‘moved’, dengyu ‘equal’

• Mei: xiangdao ‘think of’, laideji ‘be able to make it in time’, shuohua

‘speak’, jian ‘see; meet’, zhaodao ‘find’, tingjian ‘hear’, wangji

‘forget’, dedao ‘get’, kanjian ‘see’, faxian ‘discover; find’

As situations are typically realised or experienced in relation to speech

time when the reference time is unspecified it is the speech time by default (cf.

Xiao and McEnery (2004)). The distinction between bu and mei as proposed

in this article, i.e. bu negates property, volition, relationship, possibility and

necessity etc while mei only negates realisation, can also readily account for

why mei frequently negates situations with a past or present reference time, as

widely observed in the literature.

As the notion of realisation is incompatible with imperatives, mei is not

used in imperative sentences. Neither can mei be used as a potential infix. A

potential infix such as de (kan-de-qing ‘can see clearly’) or bu (kan-bu-qing

‘cannot see clearly’) essentially expresses a stative meaning by indicating

possibility. When a verb phrase with a potential infix expresses a realisation

meaning, a perfective viewpoint is taken (e.g. kan-qing-le ‘saw clearly’), and

in a negative case, mei is used (e.g. mei kan-qing ‘did not see clearly’). It is

also of interest to note that bu and mei, especially the latter, are used as a

question tag or an independent reply more frequently in speech than writing.

This is because questions and answers are an important feature of

conversations and Callhome is composed of conversations. Bu also occurs

more frequently in imperatives in the spoken corpus, which is hardly

surprising given that Callhome is a corpus of telephone consversations

between Chinese nationals living overseas and their families in China.

[FIGURE 3 NEAR HERE]

We have already noted some important differences between written and

spoken Chinese regarding negation. Now we will consider some more fine-

grained genre distinctions between bu and mei. Figure 3 shows the distribution

of bu and the adverbial uses of mei across the 16 genres (15 genres in LCMC

and the spoken genre in Callhome). For ease of comparison, the frequencies

are normalised to a common base of occurrences per 100,000 words. As can

be seen, bu is much more common than mei in all genres. This is because, as

noted earlier, mei is aspectually oriented and only negates realisation while bu

has a much wider scope of uses. Both bu and mei are generally more frequent

in the five types of fiction (K-P), humour (R), and especially speech (S),

which shows very high frequencies of both negative forms. This is

unsurprising given the interactive nature of conversation (cf. Biber (1988)),

which is common in these genres.7 In contrast, official documents (H) show

the lowest frequencies of both bu and mei while mei is also rare in academic

prose (J), trades/skills/hobbies (E), news reviews (B), news editorials (C),

religious texts (D) and popular lore (F). A common feature of these genres is

that they are essentially expository, lacking a narrative focus which is closely

associated with mei. Biographies (G) and news reportage (A) appear to be

transitory genres between narrative and expository writings. This finding is

line with the distribution pattern for aspect markers as observed in McEnery,

Xiao and Mo (2003).

4. NEGATION AND ASPECT MARKING

We noted in the earlier section that negation in Chinese is closely allied with

aspect. In addition to aspectual features such as dynamicity and telicity, the

interaction between negation and aspect marking has been a focus of previous

research. Xu (1997), for example, observes that bu is semantically

incompatible with aspect markers denoting realisation. We would like to

extend this observation and claim that mei is not compatible with the actual

aspect marker -le, though this negation marker can co-occur felicitously with

the experiential aspect marker -guo (see Xiao and McEnery (2004) for a

discussion of these viewpoint aspects). Our data shows that all situations that

can take the actual aspect marker -le or the experiential aspect marker -guo in

positive sentences can be negated by mei. In contrast, bu is never used to

negate a perfective viewpoint. Sentences in (7-9) are some corpus examples

with -le and -guo and their negative forms. As can be seen, when a sentence

taking -le is negated, -le is removed because both -le and mei mark realisation,

with the former for the positive form and the latter for the negative form.8

When a sentence with -guo is negated, there is no such redundancy because -

guo has an experiential viewpoint focus while mei only negates realisation –

an experienced situation is surely realised. Note that while both mei and bu

can be used to negate sentences like (8a), the two negative forms have

different meanings – mei negates realisation while bu negates volition.

(7) a. zhe ci gongshen dadao-le yuqi de

this Cl public:trial achieve-Asp expected DE

xiaoguo (LCMC: F)

result

‘This public trial achieved the expected results.’

b. zhe ci gongshen meiyou/*bu dadao yuqi de xiaoguo

this Cl public:trial not achieve expected DE result

‘This public trial did not achieve the expected results.’

(8) a. wo xiang ni che-le yi ge huang (LCMC: R)

I to you tell-Asp one Cl lie

‘I told a lie to you.’

b. wo meiyou/bu xiang ni chehuang

I not to you tell:lie

‘I did/will not lie to you.’

(9) a. ta qizi wo jian-guo (LCMC: P)

he wife I see-ASP

‘I have seen his wife.’

b. ta qizi wo mei/*bu jian-guo

he wife I not see-ASP

‘I have not seen his wife.’

While the interaction between negation and the actual -le and the

experiential -guo appears to be relatively straightforward, there is much

controversy surrounding the interaction between negation and the progressive

and durative aspects. For the progressive aspect marked by zai, Hsieh (2001),

Wan (2001) and Lin (2003) argue that it can only be negated by mei while Li

(1999) observes that it can be negated by either bu or mei; for the durative

aspect marked by -zhe, Li (1999), Wan (2001) and Lin (2003) argue that it can

only be negated by mei while Hsieh (2001) notes that it can be negated by

either bu or mei. In the meanwhile, Hsieh (2001) also notes that bu can be

used to negate the progressive in Taiwanese Mandarin, contrary to the claim

made by Lin (2003, 454) that nobody he knows from Taiwan accepts

examples such as Hongmei bu zai shuohua ‘Hongmei isn’t speaking’. Lin

(2003, 431) admits that some verbs (e.g. cunzai ‘exist’ and yiwei ‘mean’)

taking -zhe can be negated by bu. While both authors are native speakers of

Taiwanese Mandarin, it appears that they have totally different intuitions on

this point. The responses to our elicitation test from informants from mainland

China, Taiwan and Hong Kong also show that different people have different

intuitions. In this article, however, we will not get involved in this debate but

base our observations instead on corpora consisting of written and spoken data

produced by mainland Chinese speakers.

A number of interesting observations can be made on the basis of this

dataset. First, in relation to perfective aspects, the negative forms of

imperfective aspects are very uncommon. While there are variations in written

and spoken Chinese, our dataset as a whole indicates that the positive forms of

the actual aspect are roughly 11 times as frequent as its negative forms while

the positive forms of the experiential aspect are five times as frequent as its

negative forms.9 The positive and negative forms of imperfective aspects

display much more marked contrasts. For the progressive aspect, the

positive/negative ratio is 134 while for the durative aspect the ratio is 202.10

This means that the progressive zai and the durative -zhe rarely occur in

negative sentences. Of the 946 instances of progressive zai in LCMC and

Callhome, only seven co-occur with bu. These seven co-occurrences

themselves present an interesting phenomenon – five of them occur in double

negation sentences (e.g. 10a-b) and one appears in a rhetorical question (10c),

both of which are actually positive by nature. Only one co-occurrence was

found in a truly negative sentence (10d).

(10) a. <…> wushiwuke bu zai sinian-zhe jiaxiang (LCMC: G)

at:no:moment not Asp miss-Asp hometown

‘(These soldiers) were missing their hometowns the whole

time.’

b. Tian’anmen Guangchang zhouwei de jianyi

Tian’anmen Square around DE simply:equipped

shuiguan wu shi bu zai gongshui (LCMC: B)

tap no moment not Asp supply:water

‘The simply equipped taps around the Tian’anmen Square were

supplying water at all the time.’

c. Jin Ying bu shi, xianzai bu zai nianshu ma? (Callhome)

Jin Ying not be now not Asp study Prt

‘Isn’t Jin Ying studying now?’

d. <…> ye bu zai zhao fangzi ye bu shi sha (Callhome)

also not Asp look:for house also not be what

‘(She) is not looking for a house or something like that.’

As the LCMC and Callhome corpora only contain approximately 1.3

million words, we consulted the Mandarin News Text Corpus to find

additional examples. This much larger corpus provides a total of 50 co-

occurrences of the progressive zai with bu and two co-occurrences with mei.

As in LCMC and Callhome, the majority (over 90%) of co-occurrences with

bu appear in double negation sentences and rhetoric questions. Of the

remaining four instances, two were found in the data from China Radio

International and two in the same sentence in the People’s Daily (1993).11

While co-occurrences of the progressive zai with and bu and mei are rare,

our data also shows that both adverbs can negate the progressive as shown in

(11), though they have different focuses, as noted earlier. Clearly, the negation

in (11a) by meiyou is motivated temporally and aspectually whereas the

negation in (11b) by bu is atemporal – it simply negates the fact that the river

is still flowing. The low frequency of the progressive co-occurring with bu and

mei can be accounted for by the fact that the progressive aspect marker zai

(especially zhengzai), is inherently assertive and does not normally take a

negative adverb (cf. Hao (2003, 13)). The low co-occurring frequencies,

however, do not mean that the progressive in Chinese is negated as

infrequently. A more common way to negate the progressive is a viewpoint

shift that avoids the assertive progressive aspect marker in a negative sentence.

This can be achieved in a number of ways, as shown in (12-14). In (12) the

viewpoint is shifted to the perfective which is then negated by mei; in (13), the

combination of bu and the change-of-state (COS) le is used to negate the

current relevance (as indicated by hai ‘still’); in (14) the focus of negation is

transferred to the subject. It is also possible to negate truthfulness using bu shi

‘not be’ (e.g. 15a). A likely reply to the question in (15b) can be wo zai kanshu

‘I am reading’ or wo mei zuo shenme ‘I did nothing’.

(11) a. tamen queshi meiyou zai wei ziji kaituo (Xinhua 02/07/94)

they indeed not Asp for self absolve

‘Indeed, they have not been absolving themselves.’

b. Kushui he yi bu zai liütang, qi shangyou jin

Kushui River already not Asp flow its upper:reaches only

zai chu shan shu gongli hou bian

after out mountain several km after then

xiaoshi (China Radio 20/07/95)

disappear

‘The Kushui River is no longer flowing, as its upper reaches

dry out only several km after they flow out of the mountain.’

(12) a. bingtong zai jianqing (LCMC: A)

ailment Asp ease

‘(His) ailment is easing.’

b. bingtong meiyou jianqing

ailment not ease

‘(His) ailment is not eased.’

(13) a. xue hai zai xia-zhe (LCMC: A)

snow still Asp fall-Asp

‘It is still snowing.’

b. xue bu xia le

snow not fall COS

‘It is no longer snowing.’

(14) a. you ren zai deng wo (LCMC: F)

there:be person Asp wait I

‘Someone is waiting for me.’

b. meiyou ren zai deng wo

there:not:be person Asp wait I

‘No one is waiting for me.’

(15) a. women juedui bu shi zai chaojia (LCMC: K)

we absolutely not be Asp quarrel

‘It’s absolutely not true that we are quarrelling.’

b. ni zai zuo shenme? (LCMC: N)

you Asp do what

‘What are you doing?’

We have already noted that the durative aspect marker -zhe also co-occurs

with negative adverbs rarely. Our data shows that such co-occurrences are

typically found in formulaic expressions (16a) and V1-zhe V2 structures,

where V1-zhe acts as an adverbial modifier (e.g. 16b). If the focus of negation

falls upon current relevance, the combination of bu and COS le can be used

(17b). Like the progressive zai, -zhe is frequently omitted in negative

sentences, as shown in (18) and (19). In fact, for verbs that can occur in

negative forms either with or without -zhe (e.g. cunzai ‘exist’ and dai ‘bring’),

the those without -zhe are more frequent in our corpora. Both bu and mei can

be used to negate the durative aspect, but the temporal and aspectual

differences as noted earlier also apply in this case. For example, mei negates

the realisation of the events pan ‘expect’ in (20a) and zuan ‘hold’ in (20b)

whereas bu negates an atemporal condition which has nothing to do with

realisation in (20c).

(16) a. zhe bing bu yiwei-zhe women jiang jujue xiang renheren

this actually not mean-Asp we will refuse to anyone

chushou renhe dongxi (LCMC: C)

sell any thing

‘This does not mean that we will refuse to sell anything to

anyone.’

b. guaibude ni bu ji-zhe ba an diao qu (LCMC: P)

no:wonder you not anxious-Asp BA I transfer go

‘So that’s why you are not anxious to get me transferred.’

(17) a. ke xianzai tianshang hai piao-zhe xue (LCMC: G)

but now sky still float-Asp snow

‘But it is snowing now.’

b. xianzai tianshang bu piao xue le

now sky not float snow COS

‘It is no longer snowing.’

(18) a. renmen guanzhu-zhe ta de jinzhan (LCMC: A)

people follow:with:interest-Asp it DE progress

‘People are following its progress with interest.’

b. renmen meiyou guanzhu ta de jinzhan

people not follow:with:interest it DE progress

‘People did not follow its progress with interest.’

(19) a. qiang shang gua-zhe zihua (LCMC: P)

wall on hang-Asp calligraphy:and:painting

‘Some calligraphies and paintings are hung on the wall.’

b. qiang shang mei gua(-zhe) zihua

wall on not hang(-Asp) calligraphy:and:painting

‘No calligraphies and paintings were hung on the wall.’

(20) a. wo ke mei pan-zhe ni huilai (Callhome)

I actually not expect-Asp you return

‘I have not been expecting you to come back.’

b. shouhuoyuan jian ta shouli mei shixian zuan-zhe qian,

salesperson see he hand:in not beforehand hold-Asp money

sihu youdian qiguai (LCMC: G)

seem a:bit surprised

‘The saleswoman seemed a bit surprised when she saw that he

did not hold money in hand beforehand.’

c. ni bu kan-zhe ta dehua, hai buxing (Callhome)

you not look-Asp she Prt still won’t:do

‘It won’t do if you don’t keep an eye on her.’

5. THE SCOPE AND FOCUS OF NEGATION

A positive sentence has a definite scope of assertion, i.e. every sentential

constituent contained therein is assertive; for a negative sentence, in contrast,

its scope of negation is indefinite, because the focus of negation can fall upon

nearly any sentential constituent (cf. Dai (2000, 48)). As such, a negative

sentence, when taken in isolation, is inherently ambiguous.12 Nevertheless,

negative sentences situated in context are rarely ambiguous because the

context, including background knowledge shared by the addresser and the

addressee, must provide some clue as to which sentential constituent should be

taken into the focus of negation if the communication is to be successful. In

(21a), for example, the scope of neation is the whole sentence while the focus

of negation can fall upon, zhe ‘this’, Meiguo he duoguo budui ‘US and

multinational forces’, or zhengshi ‘confirm’, though the last of these is the

most likely reading of the sentence as it stands because the end of a sentence is

normally the semantic focus of a sentence. But when situated in context, as in

(21b), there is no alternative reading other than taking zhengshi ‘confirm’ as

the focus of negation because of the contrast present in the context, namely

fouren ‘deny’.

(21) a. zhe yi xiaoxi meiyou bei Meiguo he duoguo budui

this one news not PSV US and multinational forces

suo zhengshi

Prt confirm

‘This piece of news was not confirmed by US and multinational

forces.’

b. zhe yi xiaoxi ji meiyou bei Meiguo he duoguo

this one news both not PSV US and multinational

budui suo zhengshi, dan ye meiyou fouren (LCMC: A)

forces Prt confirm but also not deny

‘This piece of news was neither confirmed nor denied by the

US and multinational forces.’

(22) a. Jenny keneng bu shi xiongshou (LCMC: L)

Jenny possible not be murderer

‘Probably Jenny is not the murderer.’

b. ni bu zhidao! nanren bu keneng zhidao! (LCMC: K)

you not understand man bu possible understand

‘You don’t undsertand! A man cannot possibly understand

that!’

(23) a. Hanzi yu Lading zimu wei jichu de

Chinese:character and Latin character as base DE

xiwen bu tong, baohan yi ge hen da

west:language not same include one Cl very large

de zifuji (LCMC: J)

DE character:set

1) ‘Chinese is different from Western languages based on Latin

characters in that it has a very large character set.’

2) ‘Chinese is different from Western languages based on Latin

characters in that they have a very large character set.’

b. ta bu xiang yixie ren xinmu zhong de yingxiong,

he not like some person mind middle DE hero

xihuan zuijiao diao yi ke xiangyan (LCMC: L)

like mouth:corner hold one Cl cigarette

1) ‘Unlike a hero in some people’s eyes, he enjoys holding a

cigarette dangling from a corner of his mouth.’

2) ‘He is not like a hero in some people’s eyes, who enjoys

holding a cigarette dangling from a corner of his/her [the hero’s]

mouth.’

In Chinese, the scope of negation normally starts from the word

immediately following the negator and stretches to the end of the clause. Such

cases account for over 90% of our data. Meaning is changed depending upon

whether a sentential constituent is within the scope of negation or not, as

shown by the contrast in (22). Unfortunately, it is not always easy to

determine the the scope of negation. In (23a-b), for example, the scope of

negation can be taken to stop at the end of the first clause, i.e. the comma, or

to cover the second clause as well, thus resulting in multiple readings. In such

cases, the ‘correct’ reading often depends upon world knowledge and/or

presupposition absent from textual context, or even one’s expertise and

preference. In (23a), for example, as it is known that Chinese has a very large

character set with over 7,000 commonly used characters, only the first reading

is “correct”, i.e. in line with the fact; in (23b), as different readers can have

different ideas of what a hero should look like, both readings are possible for

this example. A common feature of sentences of this type is that they make a

comparison in the first clause and there is no subject in the second clause. The

empty slot for this omitted subject can be taken by either of the entities being

compared, thus causing ambiguity. Our corpora suggest that in cases like these,

the first reading is more likely to be more appropriate, as shown in (23).

For negative sentences with bu, the commonly used patterns include bu +

adjective (19%), bu + verb (16%), bu + verb + unmodified object (12%), the

potential stucture of verb + bu + complement (8%), bu + verb of the shi ‘be’

class (e.g. shi, ‘be’, shuyu ‘belong’, dengyu ‘equal’) + predicate (8%),

adverbial + verb (5%), modal verb + verb + object (5%) and modal verb +

verb (4%). These patterns account for over three quarters of the total

occurrences in LCMC. For negative sentences with mei, the commonly used

patterns include mei + verb (29%), mei + verb + complement (20%), mei +

verb + unmodified object (15%), mei + adverbial + verb (7%), mei + verb +

modified object (6%). These patterns also account for over three quarters of all

instances in LCMC. An examination of these frequent pattens shows that a

single element such as a verb or adjective following the negator is the focus of

negation (as marked in bold in 24a); the focus of negation for the pattern of a

verb followed by an unmodified object typically includes the whole predicate,

i.e. the VO structure (24b);13 when there is an attributive or adverbial modifier

(25c), a nominal or verbal quantifier (24d),14 or a complementary element

such as a resultative verb complement (24e), such elements become the focus

of negation most frequently; the potential structure (24f) and the shi structure

(24g) are normally the focus of negation; the focus of negation in the structure

of a modal verb followed by a verb (and then an unmodified object) falls upon

the verb (and also the object) (24h).15

(24) a. women haojiu meiyou jianmian le (LCMC: R)

we long not meet COS

‘We have met for a long time.’

b. cong na yihou, ta zaiye mei tan-guo duixiang (LCMC: F)

since that after he ever not talk-Asp partner

‘He has never looked for a partner for marriage since then.’

c. qishi ta meiyou zuo duibuqi wo de shiqing (LCMC: K)

in:fact he not do let:sb:down I DE thing

‘In fact, he did not do anything that let me down.’

d. guoduqi ying bu chaoguo 7 nian (LCMC: B)

transition:period should not exceed 7 year

‘The transition period must not exceed 7 years.’

e. women cai-lai-cai-qu, dou meiyou cai-zhao (LCMC: P)

we guess:over:and:over all not guess-right

‘We guessed over and over again but did not get it right.’

f. ta zuo zai di shang pa-bu-qilai (LCMC: K)

she sit on ground on climb-not-up

‘She sat on the ground, unable to get up.’

g. shangpin jingji bu shi ziji

commodity economy not be self-supporting

jingji (LCMC: J)

economy

‘Commodity economy is not self-supporting eceonomy.’

h. Qin Fang bu gan chusheng (LCMC: P)

Qin Fang not dare make:a:sound

‘Qin Fang did not dare to make a sound.’

In addition to the commonly used patterns described above, our corpora

also reveal some infrequent, yet important, negation patterns in Chinese,

which include intensifying structures marked by (lian…) ye/dou ‘even’,

pivotal structures, serial verb constructions, and imperatives with buyao ‘do

not’. For example, in (25) the intensifiers in italics make the the parts marked

in bold the informational focuses of these sentences, which naturally also

become the focuses for negation. In Chinese a pivotal structure consists of two

verb phrases with the object of the first verb functioning at the same time as

the subject of the second verb while serial verb constructions is also called ‘V-

V series’ (Chao (1968, 327)). When constructions of these two types occur in

a negative sentence, the focus of negation normally falls upon the whole

construction, as shown in (26-27). Buyao often appears in imperative

sentences. As with modal verbs, we do not include yao as the focus of

negation, because imperatives have a class meaning, namely, prohibition.

What is prohibited is more important than this class meaning (e.g. 28a).

Furthermore, there are usually modifying elements in imperatives, which must

be taken into the focus of negation (e.g. 28b).

(25) a. ta lian yi yan ye mei kan wo (LCMC: P)

she even one Cl even not look I

‘She did not even take a look at me.’

b. lian zhongwufan dou meiyou chi (LCMC: N)

even lunch even not eat

‘(He) did not eat have lunch.’

c. tamen shui ye meiyou shuohua (LCMC: G)

they who even not speak

‘Neither of them spoke.’

d. zhe shi women jia shui dou bu zhidao (LCMC: P)

this matter we home who even not know

‘No one in our family knows about it.’

(26) a. nüsheng sushe bu xiongxu nansheng

girl:student dormitory not allow boy:student

ru nei (LCMC: F)

enter inside

‘Boy students are not allowed to go into girl students’

dormitories.’

b. ta mei rang wo shuo-xiaqu (LCMC: G)

he not let I say-Asp

‘He did not let me go on.’

(27) a. meiyou zhuazhu shiji jinxing jianshexing

not grasp opportunity do constructive

duihua (LCMC: B)

dialogue

‘<…> did not grasp the opportunity to have a constructive

dialogue.’

b. zenme shutong’er ye mei jinlai bingbao

why boy:attendant ever not enter report

yi sheng? (LCMC: N)

one Cl

‘Why didn’t the boy attendant come in to report it?’

(28) a. ni buyao shengqi (LCMC: G)

you don’t get:angry

‘Don’t get angry.’

b. buyao guofen jinzhang (LCMC: N)

don’t excessively nervous

‘Don’t be too nervous.’

There are, however, exceptions to the patterns described above which

occur as a result of the contrast present in context16 as shown in examples in

(29) where the focus of negation is marked in bold and the contrastive element

is italicised. In (29a) the focus of negation is shifted from the normal position

xing ‘OK’ to the subject wo ‘I’ because of the contrast between wo ‘I’ and

nimen ‘you’; the contrast in (29b) shifts the focus to pian ‘cheat, lie’ from the

normal position of VO or O; similarly, the focus in (29c) is on jiao ‘teach’ as

opposed to xue ‘learn’ while in (29d) benzhi ‘essence’ contrasts with

xianxiang ‘appearance’; finally, the contrast between cishi ‘this moment’ and

gangcai ‘a moment ago’ results in a focus shift to the adverbial gangcai ‘a

moment ago’.

(29) a. nimen xing, ke wo bu xing a (LCMC: G)

you OK but I not OK Prt

‘You can do that, but I can’t.’

b. wo mei pian nimen, ju ju dou shi shihua (LCMC: L)

I not cheat you, CL CL all be truth

‘I did not lie to you, all that I told you was true.’

c. wo bushi lai dang laoshi de, bushi lai jiao,

I not:be come be teacher Prt not come teach

ershi lai xue (LCMC: G)

but:be come learn

‘I am not here as a teacher; I come not to teach but to learn.’

d. bu neng guang kan xianxiang bu kan benzhi (LCMC: K)

not can only look appearance not look essence

‘(You) cannot look merely at the appearance but not at the

essence.’

e. cishi ta zheng piao zai li feiyiting bu yuan de

this:moment he just float on from flying:boat not far DE

haimian shang. gangcai zenme meiyou faxian

sea:surface on just:now how not find

ta ne? (LCMC: N)

he Prt

‘At this moment he is floating on the sea not far away from the

flying boat. How didn’t we find him just a moment ago?’

6. TRANSFERRED NEGATION

Transferred negation is an issue which is closely associated with the scope and

focus of negation. It has been observed that negation in subordinate clauses

following some verbs expressing opinion (e.g. think, believe) and perception

(e.g. appear, seem) in English is conventionally transferred to main clauses (cf.

Quirk et al (1985); Biber et al (1999)). In Chinese, however, transferred

negation is considerably rarer than in English. It is uncommon even for verbs

expressing opinion and perception (e.g. renwen ‘think’, juede ‘feel; think’,

xiande ‘appear, seem’).17 There is usually a difference between transferred and

non-transferred negation. Consider the following examples:

(30) a. huilai yihou, wo juede shenme dou bu haochi (Callhome)

return after I feel what all not delicious

‘When I come back, I think no food is delicious.’

b. huilai yihou, wo bu juede shenme dou haochi

return after I not think what all delicious

‘When I come back, I don’t think all food is delicious.’

(31) a. wo zhidao ni dui bieren de mimi mei

I know you in other DE secret not:have

xingqu (LCMC: P)

interest

‘I know you do not have interests in others’ secrets.’

b. wo bu zhidao ni dui bieren de mimi you xingqu

I not know you in other DE secret have interest

‘I don’t know you have interests in other prople’s secrets.’

(32) a. Chang Zhibin juede bu name kepa le (LCMC: M)

Chang Zhibin feel not so frightful COS

‘Chang Zhibin no longer felt so frightful.’

b. Chang Zhibin bu juede name kepa le

Chang Zhibin not feel so frightful COS

‘Chang Zhibin no longer felt so frightful.’

(33) a. wo juede zhe ge bu shi hen hao de banfa (Callhome)

I think this Cl not be very good DE method

‘I think this isn’t a very good method.’

b. wo bu juede zhe ge shi hen hao de banfa

I not think this Cl be very good DE method

‘I don’t think this is a very good method.’

As shown in the translations, (30a) is an example of full negation whereas

when the negator is transferred to the main clause (30b), the rewritten sentence

becomes an instance of partial negation. Similarly, the examples in (31) also

have quite opposite presuppositions and express totally different meanings.

This is so because the position of the negator is directly relavant to the

meaning expressed, which is different from English (compare All food is not

delicious vs. Not all food is delicious). There are cases where transferred

negation does not result in such a marked contrast, as in (32-33). But in these

cases, transferred and non-transferred negations have different focuses. For

example, the focus of negation in (32a) is name kepai ‘so frightful’ whereas in

(32b) the focus is juede name kepai ‘feel so frightful’. This difference in

negation focuses produces different aspectual meanings when these sentences

take the change-of-state le, which indicates the current relevance of a situation.

Consequently, the current relevance in (32a) is bu name kepa ‘not so frightful’

whereas it is bu juede name kepa ‘no longer felt so frightful’ in (32b), as

indicated by the parts marked in italic in the two examples.

Transferred negation is essentially a type of indirect negation, which

indicates a reduced ‘strength scale’ of negation (Horn 1972). As Shen (1989, 4)

observes, the degree of negation becomes lower when the negator is further

away from what is negated. As such, (33a) expresses a stronger attitude than

(33b). The former indicates that the method is not very good while the latter

implies that the method is not very bad – if there is no alternative, it can be

used. The indirectness and reduced degree of negation also accounts for why

transferred negation typically co-occurs with verbs of opinion and perception,

which is approporiate as a politeness strategy but not effective in argumention.

7. DOUBLE NEGATION AND REDUNDANT NEGATION

Double negation is common in Chinese. Frequently used structures of this

kind include, for example, budebu ‘cannot but have to’, bunengbu ‘cannot but

have to’, wu/meiyou (…) bu ‘all (…) without exception’ and fei…(bu/mo)

‘have got to, simply must’. While it is true that double negation expresses a

positive meaning, the negation of negation often does not merely mean the

same thing as what is left when the two negators are removed (cf. Lü (1986)).

For example, the parts italicised in the (a) sentences carry a different meaning

from those in the (b) sentences in (e.g. 34-35). Similarly, bu hui/keneng bu

‘definitely’ and bunengbu ‘cannot but, simply must’ are not the same as

hui/keneng ‘probably’ and neng ‘can’ respectively. There are cases where two

negators cancel each other (e.g. 36), though double negation carries a more

emphatic force, as shown in (e.g. 37). Similar examples include bu shi meiyou

daoli (keneng, zhengyi, etc) de ‘it is not without a reason (possibility, debate,

etc)’, which mean almost the same when both negators (bu and mei) are

removed.

(34) a. Shixiong bu xiang mingtian zou, keshi diedie

Shixiong not want tomorrow leave but dad

shuo-chulai-le, jiu bu gan bu ying (LCMC: N)

say-out-Asp then not dare not agree

‘Shixiong did not want to leave the next day, but now that his

dad had said so, he had to agree.’

b. <…> jiu gan ying

then dare agree

‘(…) he dared to agree.’

(35) a. liuxuesheng yuqi wei gui bu dengyu bu

overseas:students overdue not return not equal not

aiguo (LCMC: F)

patriotic

‘Overseas students not returning within time limits does not

mean that they are not patriotic.’

b. <…> dengyu aiguo

equal patriotic

‘(…) means that they are patriotic.’

(36) a. ta meiyou liyou bu xiangxin (LCMC: M)

he not:have reason not believe

‘He had no reason not to believe.’

b. ta you liyou xiangxin

he have reason believe

‘He had reason to believe.’

(37) a. meiyou yi ge Zhongguoren bu zhidao na zhi kangkai

no:have one Cl Chinese not know that Cl fervent

de ge (LCMC: G)

DE song

‘No Chinese who does not know that fervent song.’

b. zhongguoren dou zhidao na zhi kangkai de ge

Chinese all know that Cl fervent DE song

‘All Chinese know that fervent song.’

In the structure fei…bu/mo, either a noun or verb phrase can follow fei as

shown in (38). There are two instances of the fei structure in this example,

with the first followed by a noun (dangbingde ‘soldier’) and the second by a

verb phrase (deng ta fuyuan hou ‘wait until he is demobilised’). It is of interest

to note that in the double negation structures with fei and bu, the second

negator bu can be omitted, without affecting its meaning, when a verb phrase

follows fei. For example, buke can be removed in (39a) while it is also

possible to insert buke at the end in (39b).

(38) wo fei dangbingde bu jia, erqie fei deng ta fuyuan

I not soldier not marry and not wait he demobilise

hou cai jia (LCMC: R)

after until marry

‘I will only marry a soldier but not get married until he is

demobilised.’

(39) a. wo fei yao ni gen Chen xiong bishi buke (LCMC: N)

I not want you with Chen brother compete not:can

‘I insist that you have a competition with Chen.’

b. ni bu yiding fei gaosu wo zhe jian shi (LCMC: P)

you not certainly not tell I this Cl matter

‘You don’t have to tell me about it.’

While double negation typically requires two negators that cancel each

other to produce a positive meaning, the possible omission of the second

negator in the fei…bu structure is not the only instance of redundant negation

in Chinese. Redundant negation means that a negator is not essential in

interpreting a sentence, i.e. it does not affect the meaning of the sentence. It

typically occurs in structures with some semantically negative elements.

Sentences in (40) exemplify the major structures of redundant negation in

Chinese. In all of these sentences, the negators as marked up in bold can be

removed without causing a change in meaning.

(40) a. nimen mei lai zhiqian, wo zhenshi xiahuai-le (LCMC: L)

you not come before I really frighten-Asp

‘Before you came, I was terribly frightened.’

b. haizi men juede haobu xiqi (LCMC: F)

child Pl feel very curious

‘Children feel very curious.’

c. xiaoxin bie tang-le haizi (PD 19/05/2000)

careful don’t scald-RVC child

‘Take care not to scald the child.’

d. laobaixing chadianr mei ba tian ku ta (LCMC: G)

common:people nearly not BA sky cry fall:down

‘People cried so much that the sky nearly fell down.’

e. jinhou de fazhan jincheng nanmian bu shoudao

future DE develop progress can:not:avoid not suffer

<…> deng duozhong yinsu yingxiang (PD 04/11/2000)

etc many factor influence

‘The progress of future development cannot avoid being

influenced by many factors, such as …’

f. liang ren tan-le mei yihuir, guliang bian qishen

two person talk-Asp not a:while girl then rise

gaochi (LCMC: R)

take:leave

‘The two of them talked for a while before the girl stood up to

take leave.’

Example (40a) shows that mei in the structure mei VP qian/yiqian/zhiqian

‘before not VP’ is redundant. The semantic property of mei (i.e. negating

realisation, see section 3) is compatible with the non-realisation of the

situation denoted by VP qian/yiqian/zhiqian (cf. Dai (2000, 48)), which

accounts for the felicitous co-occurrence of mei and qian/yiqian/zhiqian in this

structure as well as for the redundancy of mei. In relation to redundant

negation of this type, the alternative positive form (i.e. without mei) is nine

times as frequent. In addition, our data suggests that when the VP is a

compound verb starting with lin ‘just before’ (e.g. linshui ‘just before going to

bed’, linsi ‘just before one dies’, linxing1 ‘just before execution’, linxing2 ‘just

before leaving’, linzhong ‘just before one dies’), the redundant mei cannot be

used; when the structure VP qian/yiqian/zhiqian functions as an attributive

rather than an adverbial modifier, mei also appears dubious (e.g. ?mei qiuzhi

qian de chaoqian zhunbei ‘beforehand preparations before seeking

employment’).

In Chinese, the sequence hao + bu + adjective is an ambiguous structure

which can be analysed as haobu + adjective (e.g. haobu shufu ‘very

comfortable’) or hao + bu + adjective (e.g. hao bu shufu ‘very

uncomfortable’).18 In the first analysis, haobu is an intensifying adverb similar

to hen ‘very’ while in the second bu modifies the ensuing adjective, which as a

whole is in turn modified by hao ‘very’. As such, the first is a positive use

whereas the second is a negative one. Example (40b) shows the first usage, in

which bu is redundant. It has been noted that bu in such cases is not a negator.

Rather, it is more akin to a meaningless syllable that makes up a word, as in

bu-gan-bu-ga (i.e. ganga) ‘awkward, embarrassing’ and heibuliuqiu ‘swarthy’

(cf. Fang (1996, 67)). While the structure potentially allows two quite opposite

meanings, which can only be disambiguated in context, our data shows that

the positive use is predorminantly more frequent than the negative use in

attested data. It is important to note, however, that rongyi ‘easy’ as appears in

hao bu rongyi, when used as an adverbial, only allows the negative reading

(i.e. not easy). In fact, the positive form of the adverbial use of hao rongyi also

expresses a negative meaning. This is because hao ‘very’, unlike hen ‘very’, is

often used in an ironic expression. For example, ni hao congming can mean

‘you are not clever’ whereas ni hen congming simply means ‘you are very

clever’. This ironic use of hao actually negates what is uttered literally.

In imperatives with xiaoxin/dangxin ‘be careful, take care’ which give

warning, the negator bie/buyao ‘do not’ is redundant if what is warned againt

is out of the control of the addressee.19 As such, bie ‘do not’ in (40c) can be

removed (i.e. xiaoxin tang le haizi) without a change in meaning. Positive

forms are roughly twice as frequent as negative forms. One difference between

the two forms lies in that the latter can be systematically interpreted as two

clauses separated by a pause in speech or a comma in writing (e.g. xiaoxin, bie

tang le haizi) whereas the latter cannot (*xiaoxin, tang le haizi).

Example (40d) shows that the negator mei following chadian/chadianr

‘nearly, on the verge of’ can be redundant. Chadian/chadianr carries a

typically negative semantic prosody, though it can also occur in sentences

denoting neutral or positive situations. The negator is only redundant in

negative and neutral situations. In such cases, the variants without a redundant

negator are overwhelmingly more frequent, accounting for approxiamtely 95%

of the total occurrences. When a favourable situation is expected to occur by

the speaker, however, the negator is not redundant; rather, the positive and

negative forms express opposite meanings (cf. Zhu (1982); Shen (1999)). For

example, chadianr deshou ‘nearly succeeded’ means failing to succeed

whereas chadianr mei deshou ‘nearly did not succeed’ means a narrow

success.

Like chadian/chadianr ‘nearly, on the verge of’, nanmian ‘hard to avoid;

cannot help’ is typically used in a semantically negative context as what one

tries to avoid is rarely beneficial. The negator bu ‘not’ following nanmian

‘hard to avoid; cannot help’ can be removed or replaced with the positive form

yao/hui ‘will’ without affecting meaning (e.g. 40e). Our corpora show that

positive forms are dominant in such structures (approximately 99% of the

time). Furthermore, it is important to note that in rare cases where nanmian is

followed by words of favourable meanings, e.g. jianquan ‘healthy, perfect’ in

zhidu nanmian bu jianquan ‘The system cannot but be imperfect’, the negator

bu cannot be omitted, because one need not to avoid something favourable.

The final example of redundant negation we will examine is (40f), which

shows that redundant negation can occur when a negator co-occurs with an

adverbial quantifier indicating a small quantity. Yihui/yihuir can mean ‘a

while’ (e.g. zuo yihui ‘sit for a while’), ‘the moment’ (e.g. na yihuier ‘at that

moment’) and ‘now…now’ (e.g. yihuier ku yihuier xiao ‘cry this moment and

laugh the next’). But of these, only the first meaning is relevant to redundant

negation. Temporal duration is a flexible concept (cf. Xiao and McEnery

(2004)). As such, yihui/yihuir ‘a while’ can be conceived of as long (e.g. hao

yihuir/laoda yihuir ‘a long while’) or short (e.g. bu yihui, bu da yihuir, bu duo

yihuir, mei yihuir ‘a short while’), depending upon one’s psychological

expectation (cf. Shen (1999, 116)). When yihui/yihuir co-occurs with bu or

mei, the short duration reading is selected automatically. In this reading,

yihui/yihuir denotes a minimum quantity which is qualitatively not different

from negation, thus rendering bu or mei redundant.

8. CONCLUSIONS

This article explored negation in Chinese on the basis of corpora of written

and spoken Chinese. It has been found that in addition to negating functions,

negative adverbs in Chinese also display differences in distribution across

genres. The best characterisation of mei is that it negates the realisation of a

situation, which distinguishes it from the general purpose negator bu.

Regarding the interaction between negation and aspect marking, the actual

aspect marked by -le and the experiential aspect marked by -guo are negated

by mei while the durative aspect marked by -zhe and the progressive aspect

marked by zai rarely occur in negative sentences. Imperfective aspects of the

latter two types usually undergo a viewpoint aspect shift when they are

negated. While sentences taking zai can be negated by either bu or mei, co-

occurrences of bu and zai are typically in double negation structures or

rhetoric questions, which are essentially positive in meaning. Sentences taking

-zhe are negated more frequently by mei than bu, and -zhe is often omitted in

negative sentences unless it appears in the V-zhe V structure where V-zhe acts

as an adverbial. The scope of negation typically extends from the word

immediately following the negator to the end of a clause unless the context

provides clues that suggest otherwise. Word order is important in determining

the focus of negation, which typically falls on some modifying element which

usually follows the negator immediately or the end of a clause unless a

contrast present in context suggests otherwise. Transferred negation is

uncommon in Chinese. When it occurs, the transferred focus of negation

suggests a reduced degree of negation. Double negation is common in Chinese,

but the negation of negation often means more than two negators cancelling

each other, not only in terms of emphatic force, but in meaning as well.

Finally, redundant negation typically occurs in sentences with some element

which is inherently negative semantically. Methodologically, this study

demonstrates that a corpus-based approach can achieve a more realistic and

accurate account of linguistic phenonema than the approach that relies solely

upon one’s intuitions alone.

ACKNOWLEDGEMENTS

We are grateful to the UK ESRC for supporting our project ‘Contrasting

English and Chinese’ (RES-000-23-0553), on which the work presented in this

article is undertaken. We also thank all informants who took part in our

elicitation test.

NOTES

1 See http://www.ling.lancs.ac.uk/corplang/pdc2000/default.htm for a

description of the PDC2000 corpus.

2 Mei and meiyou used as a question tag does not have the negating function

and can be replaced by modal particle me.

3 Following Lü (1982), we consider mei/meiyou in comparative sentences as a

verb, as in Cheng Jiahe suiran meiyou Lu Xiaoman nayang boxue… ‘Although

Cheng Jiahe is not as learned as Lu Xiaoman…’ (LCMC: P).

4 Note that only adjectives denoting property can be negated. Descriptive

adjectives do not have negative forms. When mei/meiyou negate a noun (as in

meiyou ren ‘nobody’), it is used as a verb. Bu can negate a noun in special

structures like nan bu nan, nü bu nü ‘(look) neither like a man nor like a

woman’.

5 Examples are given in Chinese Pinyin. The sources of examples are

indicated in the brackets, where PD represents the People’s Daily. The

truncated parts in examples are marked by <…>. In grammatical glosses, Asp

stands for aspectual marker, BA for the marker for preposed object ba, Cl for

classifier, COS for change-of-state le, DE for structural paticle de of various

forms, Pl for the plural suffix men, Prt for particle, PSV for passive, and

finally RVC for resultative verb complement.

6 According to Shi (2001), when they negate an adjective, bu differs from mei

in that the former negates the extent of the property denoted by the adjective

whereas mei negates the existence of the property. For example, tian hai bu

liang means that the sky is light, but the extent is not high (i.e. not light

enough to be considered as daytime), while tian hai mei liang means that the

sky is not light at all, because the realisation of the situation tian liang is

negated.

7 The relative low frequency of mei in martial arts fiction (N) is compensated

for by its highest frequency of wei, a more archaic negative adverb with a

similar function (see section 2).

8 One might argue that mei can co-occur with -le as in meiyou huang le

xinshen, luan le shoujiao ‘did not get flustered or thrown into confusion’

(LCMC: A) and mei wang le dai xinxian cuyu ‘did not forget to bring freshly

pickled fish’ (LCMC: G). In examples like these, however, le is a resultative

verb complement which is equivalent to liao ‘finished’ or diao ‘detached’.

9 As mei does not co-occur with -le, the positive/negative ratio for the actual

aspect was computed on the basis of the frequencies of -le and mei as an

adverb modifying stative, dynamic and modal verbs in Figures 1-2.

10 The count for -zhe does not include V1-zhe V2 structures as V1-zhe

functions as an adverbial modifier.

11 Some of our informants suggest that the two instances in the broadcast

script from China Radio International are very likely to be transcription errors.

The two examples are (11b) and dan zhe jiujing shi “yiju” haishi “yiwu”,

renmen de yilun bu zai (or buzai) jixu ‘But whether this is ‘a magnificent act

undertaken for the public good’ or an ‘obligation’, people are no longer

commenting (or people’s comments no longer continue)’.

12 A positive sentence is also likely to allow multiple readings arising from

different semantic and/or informational focuses in different contexts.

13 Note that the VO structures also include object clauses (e.g. conglai mei

tingshuo-guo jiashi feiji de ren shi zhan-zhe de ‘It is unheard of that someone

stands when piloting a plane’, LCMC: E) and preposed objects (e.g. shenme

xiansuo dou meiyou faxian ‘No clue was found’, LCMC: M).

14 When quantifiers such as yi ‘one’ and ban ‘half’ take bu or mei, what is

negated is in effect the situation per se (e.g. mei wo de anpai, dongxi ban ge

bu neng dong, qian yi fen bu neng hua ‘Without my arrangement, nothing

should be moved, nor a penny should be spent’, LCMC: L). Pronouns like

shenme ‘what’ and renhe ‘any’ have a similar effect. In contrast, when a

quantifier referring to a quantity greater than one takes a negative adverb,

what is negated is the extent specified by the quantifier.

15 It has been argued in the literature that the modal verb should be included in

the focus of negation. They should not, in our view, because while modal

verbs have some class meaning (e.g. possibility, volition, etc) on their own,

the focus of negation usually occurs in what follows the modal verb, as shown

in renmen bu gan zai zhe zhong shu xia xiuxi ‘People do not dare to rest under

trees of this kind’ (LCMC: D).

16 The contrast can be indicated by the tone and stress in speech. But we will

not discuss such spoken features in this article as our corpora are not

prosodically annotated.

17 What is expressed by verbs of perceptions in English can also be achieved

by adverbs such as sihu ‘appear, seem’, haoxiang ‘look like’, and keneng

‘likely’. Sihu and haoxiang do not take a negator (compare sihu/haoxiang bu

hui xiayu ‘It doesn’t look like it’s going to rain’ vs. *bu sihu/haoxiang hui

xiayu) while bu keneng ‘impossible’ and keneng bu ‘probably not’ mean

differently.

18 The slot taken by adjective in the formula can be replaced by an adjectival

verb such as aonao ‘fret’, houhui ‘regret’, shengshi ‘get angry’, xianmu

‘admire, envy’ and zhendong ‘shock’.

19 Note that when an imperative does not give warning, or when the addressee

has control over what is warned against, the positive and negative forms mean

differently (compare xiaoxin bie guo malu ‘Be careful not to cross the road’ vs.

xiaoxin guo malu ‘Cross the road carefully’, Dai (2004)).

REFERENCES

Bai, Quan (2000) “‘Bu’, ‘Mei(you)’ Jiaoxue he Yanjiu shang de Wuqu [Error

Zones for Bu and Mei(you) in Teaching and Research],” Yuyan Jiaoxue yu

Yanjiu 2000/3, 21-25.

Biber, Douglas (1988) Variation across Speech and Writing. Cambridge

University Press, Cambridge.

Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward

Finegan (1999) Longman Grammar of Spoken and Written English,

Longman, London.

Chao, Yuanren (1968) A Grammar of Spoken Chinese, University of

California Press, Berkeley.

Dai, Yaojing (2000) “Shilun Xiandai Hanyu de Fouding Fanchou [On the

Category of Negation in Modern Chinese],” Yuyan Jiaoxue yu Yanjiu

2000/3, 45-49.

Dai, Yaojing (2004) “Shishuo ‘Rongyu Fouding’ [A Discussion of

‘Redundant Negation’],” Xiuci Xuexi 122, 3-6.

Fang, Xujun (1996) “Xi Hai/haobu + Xingrongci de Tongyi Xianxiang [An

Analysis of the Synonymous Phenomenon of Hao/haobu + Adjective],”

Shanghai Shifan Daxue Xuebao 1996/3, 65-68.

Hao, Leihong (2003) Xiandai Hanyu Fouding Fuci Yanjiu [A Study of

Negative Adverbs in Modern Chinese], MA dissertation, Shoudu Shifan

Daxue, Beijing.

Horn, Laurence (1972) On the Semantic Properties of Logical Operators in

English, PhD dissertation, University of California, Los Angeles.

Hsieh, Miao-Ling (2001) Form and Meaning: Negation and Question in

Chinese, PhD dissertation, University of Southern California, Los Angeles.

Hundt, Marianne, Andrea Sand and Rainer Siemund (1998) Manual of

Information to Accompany the Freiburg-LOB Corpus of British English,

University of Freiburg, Freiburg.

Krishnamurthy, Ramesh (2000) “Collocation: from Silly Ass to Lexical Sets,”

in Chris Heffer and Helen Sauntson (eds.) Words in Context: A Tribute to

John Sinclair on His Retirement, University of Birmingham, Birmingham.

Li, Mei (1999) Negation in Chinese, PhD dissertation, University of

Manchester, Manchester.

Li, Charles and Sandra Thompson (1981) Mandarin Chinese, University of

California Press, Berkeley.

Lin, Jo-Wang (2003) “Aspectual Selection and Negation in Mandarin

Chinese,” Linguistics 41/3, 425-459.

Liu, Linda (1997) Guo Min Ke Yu de Dongtai Wenfa Tixi ji Dongtai Ci de

Shangjia Dongmao Yuyi [The Mood System in Mandarin, Southern Min

and Hakka, and the Aspect Meaning Posted on the Modals], the Crane

Publishing Co., Ltd., Taipei.

Lü, Shuxiang (1982) Xiandai Hanyu Babai Ci [800 Words in Modern

Chinese], Shangwu Yinshuguan, Beijing.

Lü, Shuxiang (1986) “Guanyu Fouding de Fouding [About the Negation of

Negation],” Zhongguo Yuwen 1986/1.

McEnery, Anthony and Zhonghua Xiao (2002) “Domains, Text Types, Aspect

Marking and English-Chinese Translation,” Languages in Contrast 2/2,

51-69.

McEnery, Anthony, Zhonghua Xiao and Lili Mo (2003) “Aspect Marking in

English and Chinese: Using the Lancaster Corpus of Mandarin Chinese for

Contrastive Language Study,” Literary and Linguistic Computing 18/4,

361-378.

McEnery, Anthony, Zhonghua Xiao and Yukio Tono (2005) Corpus-Based

Language Studies: An Advanced Resource Book, Routledge, London.

Nie, Renfa (2001) “Foudingci ‘Bu’ yu ‘Meiyou’ de Yuyi Tezheng jiqi Shijian

Yiyi [The Semantic Features and Meanings of Time on Two Negatives: Bu

and Meiyou],” Hanyu Xuexi 2001/1, 21-27.

Qu, Yinjie (2002) Xiandai Hanyu Fouding Fuci “Bu” he “Meiyou”, MA

dissertation, Huanan Shifan Daxue, Guangzhou.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik (1985)

A Comprehensive Grammar of the English Language, Longman, London.

Seuren, Pieter (1998) Western Linguistics: A Historical Introduction.

Blackwell, Oxford.

Shen, Jiaxuan (1989) “‘Panduan Yuci’ de Yuyi Qiangdu [The Semantic

Strength of ‘Predicative Words’],” Zhongguo Yuwen 208, 1-8.

Shen, Jiaxuan (1995) “Youjie yu Wujie [Bounded and Unboundedness],”

Zhongguo Yuwen 1995/5, 367-380.

Shen Jiaxuan (1999) Buduicheng yu Biaoji Lun [Asymmetry and a Theory of

Markedness], Jiangxi Jiaoyu Chubanshe, Nanchang.

Shi, Yuzhi (2001) Kending yu Fouding de Duicheng yu Buduicheng

[Symmetry and Asymmetry of Assertion and Negation], Beijing Yuyan

Wenhua Daxue Chubanshe, Beijing.

Wan, Ying (2001) Fouding Fuci “Bu” he “Mei(you)” de Bijiao Yanjiu [A

Comparative Study of the Negative Adverbs “Bu” and “Meiyou”], MA

dissertation, Huazhong Shifan Daxue, Wuhan.

Xiao, Zhonghua and Anthony McEnery (2004) Aspect in Mandarin Chinese:

A Corpus-Based Study, John Benjamins, Amsterdam.

Xu, Ding (1997) Functional Categories in Mandarin Chinese, Holland

Institute of Generative Linguistics, Leiden.

Zhu, Dexi (1982) Yufa Jiangyi [Lecture Notes on Grammar], Shangwu

Chubanshe, Beijing.

TABLES AND FIGURES

Table 1 Genres covered in LCMC

Code Genre No. of samples Proportion

A Press reportage 44 8.8%

B Press editorials 27 5.4%

C Press reviews 17 3.4%

D Religion 17 3.4%

E Skills, trades and hobbies 38 7.6%

F Popular lore 44 8.8%

G Biographies and essays 77 15.4%

H Miscellaneous (reports, official

documents)

30 6%

J Science (academic prose) 80 16%

K General fiction 29 5.8%

L Adventure fiction 24 4.8%

M Science fiction 6 1.2%

N Adventure fiction 29 5.8%

P Romantic fiction 29 5.8%

R Humour 9 1.8%

Total 500 100%

Table 2 Proposals of major distinctions between bu and mei

Criteria Bu Mei

Subjective vs. objective Subjective Objective

Reference time Past, present, future Past and present

Stative vs. dynamic Stative Dynamic

Completion vs. non-completion Non-completion Completion

Telic vs. atelic Atelic Telic

Bounded vs. unbounded Unbounded Bounded

Realis vs. irrealis Irrealis Realis

Usage

tag/replystative

modalinfix

imperativedynamic

advadj

Perc

enta

ge100

80

60

40

20

0

Form

bu

mei/meiyou

Figure 1 Bu and mei/meiyou in LCMC

Usage

tag/replystative

modalinfix

imperativedynamic

advadj

Perc

enta

ge60

50

40

30

20

10

0

Form

bu

mei/meiyou

Figure 2 Bu and mei/meiyou in Callhome

Genre

SRPNMLKJHGFEDCBA

Freq

uenc

y pe

r 100

,000

wor

ds3000

2000

1000

0

Form

mei/meiyou

bu

Figure 3 Distribution of negative adverbs across 16 genres